Hit enter to search or ESC to close

How Can a Self-Driving Car Operation Improve Ride-Hail and Goods Delivery?

As the global economy grapples with an ever-evolving pandemic, we’re starting to see consumer trends in ride hailing and goods delivery that look favorably upon the opportunity presented by the deployment of autonomous vehicles. 

Where goods delivery is concerned, the pandemic has accelerated consumer usage and increased openness to products delivered by autonomous vehicles, according to a recent McKinsey report. Already, we’re seeing retailers and logistics companies looking at new ways to structure their storefront and warehousing locations to facilitate post-pandemic delivery and pick up.  And according to companies like Lyft, we’re also seeing a return to ride-hailing, indicated by an increase of 3.6 million active riders in the second quarter of 2021 compared to the previous quarter. 

This all adds up to renewed momentum for self-driving technology. That’s why our test fleet is operating ride-hail and delivery scenarios in six cities across the country every day, enabling us to partner with Lyft and Ford to launch an autonomous ride-hail service available to the public later this year. Building off of our large urban-testing footprint, with international expansion to Munich and Hamburg coming soon, our product team is refining our technology daily to address the nuances of both types of services—delivery and ride-hail. 

Solving for Both

The core functionality behind the Argo self-driving system (SDS) applies to both goods delivery and ride-hailing. The ability to autonomously navigate through a city safely; the ability to plan routes; to understand the rules, regulations, and idiosyncrasies of different road types; to respond safely to all actors on the road—all are crucial whether we’re transporting people or goods.

But since our earliest days as a company, we’ve tested our fleet in a launch-intent manner. This is because testing in the real world often highlights non-obvious problems for a product, and it helps us to expand and enrich our virtual simulations. Real-world testing has enabled us to conduct pilot programs with partners, further enhancing our SDS. It has led to the creation of our cloud-based dispatch system ArgoWatch, our own internal ride-hail app, which allows us to test daily operations in preparation for seamless integration with any ride-hail service’s systems. We also operate a “phantom fleet” that simulates everything from a single car on a set of routes to 1 million virtual vehicles simultaneously servicing multiple large metropolitan areas. 

These are just some of the critical puzzle pieces that allow the Argo SDS to master the diversity of scenarios encountered in goods and people-moving services. Some of the differences between ride-hailing and goods delivery are obvious, but many are more subtle. Below I outline a handful of the differences that we’ve identified—and solved for—including the ways trips are structured; how to safely stop or park for pick-ups, drop-offs, and deliveries; and the nature of wait times for passengers and for deliveries.  

Routing Trips

We began by exploring something fundamental to both ride-hail and delivery services: routing trips. When people use ride-hailing services, they expect to be able to travel “point-to-point,” from their current location to an address of their choice. So we built an SDS capable of delivering on that expectation within our area of operation, which can easily plug into the operations of a ride-hail company, like Lyft, in that area. Our test fleet actively mimics a ride-hail experience by randomizing a sequence of destinations throughout each day. We build flexibility into our routing services for ride-hail, including the ability to pull over at the nearest safe stopping point if a passenger requests it by cancelling the ride, and the ability to add an additional stop en route should a passenger’s plans change. 

For goods delivery, in addition to the point-to-point routing model (which is used by some goods delivery services), we also support a predetermined route model—meaning the self-driving vehicle will stop in identified destinations in a fixed order. As a result, we’re building a system that can optimize for maximum efficiency across a variety of goods-moving use cases. 

Picking a Parking Spot 

Figuring out how—and where—a self-driving vehicle can safely stop on busy city streets as it awaits or delivers its “passenger” (whether people or goods) is central to both of these types of services, and drastically different from the experience of driving in the suburbs, where there are likely defined stopping zones or parking lots. 

A unique aspect of operating an SDS in the complex urban core of a city like Miami is that ride-hailing services, like the one we’re powering for Lyft, must be capable of in-lane stopping (safely coming to a brief stop in the lane) or pulling over to the curb if space is available. In the event neither of those options are safe, the SDS needs to notify the rider that the vehicle is finding a safe stopping place nearby—and provide walking directions to get there. 

For goods delivery, the ability to stop temporarily is even more important, especially since multiple deliveries may take place during a single stop, such as at an apartment building. After alerting the customers that their package has arrived, the self-driving vehicle will need to find a parking spot where it can remain for a longer period of time. 

Our recent collaboration with The Education Fund to deliver groceries and supplies to Miami families not only benefited those families, but was particularly helpful for training our SDS to respond to different parking conditions. Over the course of eight weeks, we gained crucial insight into how the presence of curb-parking, different amounts of traffic, stops of varying lengths, and a variety of lane configurations impact stopping safety. We also learned that by annotating our high-definition maps of cities with information about access to parking on certain streets and apartment complexes ahead of time, we can prepare our self-driving vehicles for the safest possible stopping options. We also have the ability to embed dynamic information in our maps—like street-cleaning dates and parking based on days of the week—that position our fleet for goods delivery under the unique conditions across each city.  

Making Pick-ups and Drop-offs

Finally, we’ve built an SDS capable of stopping for different lengths of time at pick-up and drop-off—a crucial attribute for both services. 

For ride-hailing pick-ups, even though it’s nice to think that every passenger will be ready the instant we arrive, people don’t always work this way. The system we are building accounts for passengers who are running late. In the event of a delay, our system can stay where we are (assuming there’s a safe place to pull over) or find a nearby place to park and notify our passenger of the vehicle location. Ride-hailing drop-offs are typically quicker because passengers are eager to get out promptly at their destination, whereas it can take longer to unload a package at its final destination.

For goods delivery pick-ups, the process might seem simple—the self-driving vehicle just pulls into a parking lot and up to a loading facility and receives the goods. But there are a number of different shipping environments that are encountered in the movement of goods, like warehouses, retailers, and restaurants. Because of this variety, we’re improving our system’s ability to navigate the loosely structured nature of parking lots, where lane lines may not exist or are often disregarded, and where pedestrians abound. 

With our new long-range lidar technology, Argo Lidar, we have unlocked the ability to operate at higher speeds. As a result, we’re now testing on highways, which will enable us to connect urban and suburban neighborhoods and traverse entire large metropolitan areas. This capability allows us to support ride-hail services like Lyft to and from airports, hotels, casinos, stadiums, and all other special-event facilities where there is often the highest demand. For this purpose, we’ve built in queuing capabilities that can be used at predetermined pickup and drop-off areas.

We believe the early successes in our goods-delivery and ride-hailing testing validate our scalable approach to developing a self-driving system, and we see the post-pandemic market trends as indications that the world is more ready than ever to embrace our technology. Thanks to our testing in multiple cities in the U.S. and, soon, in Germany, we’ve gained critical insights into the ways that solving for goods delivery complements, and often overlaps with, building a product for moving people. 

All of this wouldn’t be possible without ArgoWatch, a crucial piece of digital infrastructure that is the unsung hero in Argo’s ultimate quest: to bring autonomous ride-hailing and goods delivery to life.  

Choose your lane

How Autonomous Vehicles Distinguish Between Bicycles and People Who Ride Them

How Autonomous Vehicles Distinguish Between Bikes and People

When it comes to how autonomous vehicles see the world, humans come first, literally. Autonomous vehicles (AVs), like the kind operated by Pittsburgh-based Argo AI, use Machine Learning to detect and classify the objects in their surroundings, identifying people...
Why The League of American Bicyclists is optimistic about autonomous vehicles

Why a Leading Cycling Advocacy Group Is Optimistic About Autonomous Vehicles

As autonomous vehicle use grows, AV companies and the League of American Bicyclists are collaborating on how to ensure cyclists and motorists can share the roads safely, even if the “motorist” is artificial intelligence software. As part of the...

Self-Driving Is Arriving Right On Time. Just Like Ice Cream Did

Seven years ago, I was a self-driving skeptic. Not of the technology. Of all the “experts” promising autonomous vehicles would be everywhere by 2020. You didn’t need to be Nostradamus to know that was ridiculous. All you needed was...
Illustration of a futuristic parking deck turned into a mixed-use space, with AVs driving by

How Autonomous Vehicles Could Help Transform Parking Lots

Researchers say it’s likely that autonomous vehicles (AVs) can help reduce the need for parking lots, opening more room for grass and trees and other elements of nature. It may not seem like it when you’re circling the block...
An illustration of an Argo autonomous vehicle in teal against a backdrop of buildings, a bicyclist, and research papers

7 Big Breakthroughs From Argo Research at CVPR 2022

The 2022 Conference on Computer Vision and Pattern Recognition (CVPR 2022) is nearly here. Thousands of computer scientists, software engineers, and researchers from around the globe will gather in New Orleans to review and discuss their latest work in...

Researchers Predict the Future With Lidar Data

Researchers at Carnegie Mellon University’s Argo AI Center for Autonomous Vehicle Research, a private-public partnership funded by Argo for advancing the autonomous-vehicle (AV) field, say they have come up with a way to use lidar data to visualize not...

Must Reads